In this paper, we study the throughput of multiclass networks featuring several types of flexible servers as well as general constraints both on the number of servers having the same skills and on the number of servers allowed at each station. Each job class is characterized by its processing station, and each station is characterized by the amount of work assigned to that station upon job arrival. Servers may have different skills and efficiencies. We propose a simple method calculating an upper bound for the maximum network throughput achievable with static server allocation.
